Файл:Neural Networks and Physical Systems with Emergent Collective Computational Abilities Hopfield82.pdf

ABSTRACT

Computational properties ofuse to biological organisms or to the construction of computers can emerge as collective properties of systems-having a large number of simple equivalentcomponents (orneurons). Thephysicalmeaningofcontent-addressable memory is described by an appropriate phase space flow of the state of a system. A model of such a system is given, based on aspects ofneurobiology but readily adapted to integrated circuits. The collective properties ofthis model produce a content-addressable memory which correctly yields an entire memoryfrom any subpartofsufficient size. The algorithm for the time evolution ofthe state ofthe system is based on asynchronous parallel processing. Additional emergent collective properties include some capacity for generalization, familiarity recognition, categorization, error correction, and time sequence retention. The collective properties are onlyweakly sensitive to details ofthe modeling or the failure ofindividual devices.